Re: Re: Parallel runner & Eclipse?

"Chris [email protected] [junit]" <[email protected]> Mon, 10 Oct 2016 12:35:23 +0000 (UTC)
Newsgroups gmane.comp.java.junit.user
Message-ID <[email protected]>
Hi,

- We limit the number of threads on the pool.
- I tried with some debug prints as well as some profiling and the number o=
f threads/pools look right
- I get the same behavior without the custom runner

Then your comments got me thinking a bit, so I went back to the investigati=
ons we did some time back and I think that I have to update what I said pre=
viously. Must have been some confusion on our side when we analyzed it prev=
iously.

- Many threads isn't really the problem, it is just one thread that seems t=
o get any significant CPU time. It rather seems that it is "overflowing" to=
 occupy more than one core, which surprises me a bit considering the applic=
ation under test is completely single-threaded. Tried to reproduce it by ma=
king some dummy code, but that stayed at a load of about "one core", so I'm=
 confused to say the least.
- I played around with the garbage collection, but that doesn't change the =
picture.

So it is probably that I don't understand how the JVM works rather than an =
issue with junit or Eclipse :)

Thanks for the help.

/Chris

--------------------------------------------
On Sun, 10/9/16, [email protected] [junit] <[email protected]> wrote:

 Subject: [junit] Re: Parallel runner & Eclipse?
 To: [email protected]
 Date: Sunday, October 9, 2016, 6:09 PM
=20=20=20=20=20=20=20
       It's hard to suggest more without
 seeing the code.
 Does the problem happen when
 you run tests in Eclipse that do not use your custom runner?
 Do=C2=A0you limit the
 number of threads when you configure the `ThreadPoolExecutor`? Did you try=
 adding the print
 statements that I suggested?
=20
         Posted by: [email protected]=20=20=20=20=20=20=20=20


------------------------------------
Posted by: Chris <[email protected]>
------------------------------------